Ethics

Accountability is the obligation to explain, justify, and take responsibility for actions and decisions, particularly in AI systems.

AI Safety focuses on ensuring artificial intelligence systems operate reliably and ethically, minimizing risks to humans and society.

Algorithmic fairness ensures that algorithms treat individuals and groups equitably, minimizing bias and discrimination.

Alignment in AI refers to ensuring that AI systems' goals and behaviors are consistent with human values and intentions.

Artificial Superintelligence (ASI) refers to AI that surpasses human intelligence in all aspects.

Bias in AI refers to systematic errors in algorithms that lead to unfair outcomes based on attributes like race or gender.

Bias mitigation refers to techniques used to reduce unfair bias in AI systems.

Corrigibility refers to an AI's ability to accept corrections and updates while remaining aligned with user intentions.

Capabilities of AI that pose risks to safety, privacy, or ethical standards.

Data Attribution refers to the process of identifying the source and ownership of data used in AI models.

Deceptive Alignment refers to a situation where an AI's goals appear aligned with human values but actually lead to unintended consequences.

Deepfake technology uses AI to create realistic fake audio and video content.

Demographic parity ensures equal outcomes across different demographic groups in AI decision-making.

Artificial intelligence technologies that can be used for both beneficial and harmful purposes.

Equalized Odds is a fairness criterion ensuring equal true positive and false positive rates across different groups.

Ethical AI refers to the design and implementation of artificial intelligence systems that align with moral values and societal norms.

The EU AI Act is a regulatory framework aimed at ensuring safe and ethical AI development and use in the European Union.

Existential risk refers to threats that could end human civilization or permanently curtail its potential.

Fairness in AI refers to the impartial treatment of individuals or groups in algorithmic decision-making.

Group Fairness ensures that AI systems treat different demographic groups equitably.

Individual fairness ensures similar individuals receive similar treatment in AI systems.

Misalignment refers to the discrepancy between an AI system's goals and human values or intentions.

A Model Card is a document that provides detailed information about an AI model, including its intended use and performance metrics.

Moral reasoning is the process of determining right from wrong in ethical dilemmas.

Offensive AI refers to artificial intelligence systems used to conduct harmful or malicious activities.

Outer Alignment refers to ensuring that an AI's goals align with human values and societal norms.

Predictive Parity ensures that a model's predictions are equally accurate across different groups.

A Responsible AI Framework ensures ethical development and deployment of AI technologies.